----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/28972/#review64824 -----------------------------------------------------------
Patch looks good. Two questions: 1. Are the changes applicable to other engine (harmless at least)? If so, maybe we should make the changes in trunk. 2. Adding a new alias to alias map seems a little hacky. Is it possible that we just replace the old alias with the new alias w/o keeping a mapping between them? ql/src/java/org/apache/hadoop/hive/ql/optimizer/AbstractBucketJoinProc.java <https://reviews.apache.org/r/28972/#comment107613> Is this spark specific, or also applicable in general? If it doesn't break things, we should just remove the "spark" check. ql/src/java/org/apache/hadoop/hive/ql/optimizer/MapJoinProcessor.java <https://reviews.apache.org/r/28972/#comment107612> This changes seems not required for Spark, right? - Xuefu Zhang On Dec. 11, 2014, 11:07 p.m., Jimmy Xiang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/28972/ > ----------------------------------------------------------- > > (Updated Dec. 11, 2014, 11:07 p.m.) > > > Review request for hive and Xuefu Zhang. > > > Bugs: HIVE-9081 > https://issues.apache.org/jira/browse/HIVE-9081 > > > Repository: hive-git > > > Description > ------- > > Updated the posToAliasMap in case any alias is changed. > > > Diffs > ----- > > ql/src/java/org/apache/hadoop/hive/ql/optimizer/AbstractBucketJoinProc.java > 70c23a6 > ql/src/java/org/apache/hadoop/hive/ql/optimizer/BucketJoinProcCtx.java > 911b1c8 > ql/src/java/org/apache/hadoop/hive/ql/optimizer/MapJoinProcessor.java > 773c827 > > ql/src/java/org/apache/hadoop/hive/ql/optimizer/spark/SparkMapJoinOptimizer.java > 7a716a9 > > Diff: https://reviews.apache.org/r/28972/diff/ > > > Testing > ------- > > With this patch, the auto_sortmerge_join_9.q is ok with HIVE-8639. Otherwise, > it fails. > > > Thanks, > > Jimmy Xiang > >